Uncertain databases. Scanning the research frontier

Authors:

  • Ronnie Johansson
  • Susanna Nilsson
  • Björn Pelzer

Publish date: 2014-12-31

Report number: FOI-R--4041--SE

Pages: 59

Written in: English

Keywords:

  • Probabilistic database
  • Uncertain database
  • Database queries
  • Data
  • mining
  • Performance
  • Tools

Abstract

The interest in applications which manage uncertain data is constantly increasing, including GPS positioning, Internet information extraction, autonomous navigation and decision making. Yet, off-the-shelf commercial databases lack a standardized way of dealing with uncertain information, hence the interest in database systems tailored to manage uncertainty. The research field Uncertain databases (or Probabilistic databases) studies methods to attach various types of uncertainty to data, define appropriate database operations, solve scalability problems, and integrate with data analysis algorithms. This report briefly introduces the properties of standard relational databases (which form a basis for most uncertain databases) and compare to general uncertain databases (the difference concentrated to the database manager). We then, discuss and survey in detail approaches to queries and data mining algorithm for uncertain databases. We continue to discuss performance issues and survey existing (maintained or not) research prototypes